Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Thu, 10 Nov 2022 16:07:14 -0500
From:   Peter Xu <peterx@redhat.com>
To:     Mike Kravetz <mike.kravetz@oracle.com>
Cc:     linux-mm@kvack.org, linux-kernel@vger.kernel.org,
        Naoya Horiguchi <naoya.horiguchi@linux.dev>,
        David Hildenbrand <david@redhat.com>,
        Axel Rasmussen <axelrasmussen@google.com>,
        Mina Almasry <almasrymina@google.com>,
        Nadav Amit <nadav.amit@gmail.com>,
        Rik van Riel <riel@surriel.com>,
        Vlastimil Babka <vbabka@suse.cz>,
        Matthew Wilcox <willy@infradead.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Wei Chen <harperchen1110@gmail.com>, stable@vger.kernel.org
Subject: Re: [PATCH v8 1/2] hugetlb: don't delete vma_lock in hugetlb
 MADV_DONTNEED processing
Message-ID: <Y21oAisW6AswiNIv@x1n>
References: <20221108011910.350887-1-mike.kravetz@oracle.com>
 <20221108011910.350887-2-mike.kravetz@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20221108011910.350887-2-mike.kravetz@oracle.com>
Precedence: bulk

Maybe it'll also be good if we split this into a few smaller ones?

For example:

On Mon, Nov 07, 2022 at 05:19:09PM -0800, Mike Kravetz wrote:
> Address issue by:
> - Add a new zap flag ZAP_FLAG_UNMAP to indicate an unmap call from
>   unmap_vmas().  This is used to indicate the 'final' unmapping of a vma.
>   When called via MADV_DONTNEED, this flag is not set and the vm_lock is
>   not deleted.

One patch to fix the real thing (patch 2).

> - mmu notification is removed from __unmap_hugepage_range to avoid
>   duplication, and notification is added to the other calling routine
>   (unmap_hugepage_range).

One patch to fix the mmu notifier (patch 1).

> - notification range in zap_page_range_single is updated to take into
>   account the possibility of hugetlb pmd sharing.

Part of patch 2.

> - zap_page_range_single renamed to __zap_page_range_single to be used
>   internally within mm/memory.c
> - zap_vma_range() interface created to zap a range within a single vma.

Another patch 3 for the two.

> - madvise_dontneed_single_vma is updated to call zap_vma_range instead of
>   zap_page_range as it only operates on a range within a single vma

Part of patch 2.

Then patch 1 & 2 will need to copy stable, 3 won't need to.  Patch 2 in
this series will be patch 4.  Not sure whether that looks cleaner.

Mike, sorry again if this is too late as comment, feel free to go with
either way you think proper.

Thanks,

-- 
Peter Xu